Delivery destinations

After a pipeline collects and optionally validates a file, it can deliver the result to one or more destinations. You can configure different destinations for pass, review, and fail outcomes.

Supported destinations

SFTP / FTP

Write the file to a folder on a remote server. Specify the credential and target path.

S3-compatible storage

Write the file to a bucket on Amazon S3, MinIO, or a similar service. Specify the credential, bucket, and key prefix.

HTTP API

Send the file contents to a REST endpoint using POST, PUT, or PATCH. You can map fields, add constants, and choose whether to wrap records.

Database

Insert rows into a table. Supported modes are insert, upsert, replace, and append. For upsert, specify the key columns.

Lookup dataset

Load the output into a reference dataset that standardizers can use for enrichment or validation.

Outcome-based delivery

Each pipeline can have up to three delivery actions:

• on_pass: used when the file passes validation.
• on_review: used when the file is flagged for review.
• on_fail: used when the file fails validation.

If you do not configure a destination for an outcome, the file is only staged internally and is not delivered externally.

File formats

Pipeline output is written as CSV or TSV before delivery. Database and HTTP API destinations consume rows from this delimited file.

Example use case

A typical pipeline might:

1. Collect an invoice file from a vendor SFTP server.
2. Validate it against an invoice contract.
3. On pass, deliver the file to your internal S3 bucket.
4. On fail, deliver the file to a quarantine SFTP folder for manual review.